138 COLOUR VISION 



had the smallest, and the last, green, the largest percentage of right 

 cases. To test if this were due to general improvement from increased 

 practice rather than specific qualitative differences the " repeat bright- 

 ness value series " was given. In this series all four colours were re- 

 tested with additional precautions. (3) " Hue discrimination series." 

 (a) Coloured cards and equivalent greys were exhibited in pairs : and 

 (6) two coloured cards were shown simultaneously, such that (o) the 

 pairs were equal in brightness and saturation ; {/3) the standard colour 

 was slightly or obviously darker than the other ; (7) the standard colour 

 was slightly or obviously lighter than the other. (4) " Grey difference 

 threshold series." (5) " Colour threshold series." 



Enough has been described to show that the tests were exhaustive, 

 and elaborate efforts were made to determine the nature of the dis- 

 crimination evinced. Perhaps the most serious objection which can be 

 brought against the experiments is their extreme elaboration. Anyone 

 who has had experience of testing the colour vision of untrained 

 human observers knows the frequency with which luminosity and colour 

 matches are confused. How much more this source of error is likely 

 to be present in dogs needs only to be mentioned. It was allowed for 

 to a certain extent by a graduated method of marking. As Miss Smith 

 says, " In view of their relative difficulty, sensory discrimination tests 

 on the higher vertebrates might aptly be compared to complicated 

 intellectual tests on man, in which the subject, unable to grasp completely 

 the point at issue, has to rely on a series of sporadic gleams." 



Only a summary of the chief results can be given here, the reader 

 being referred for details to the admirably thought out original paper. 

 In favour of the view that dogs are only able to perceive differences of 

 brightness are (1) the result of the " colour preference series " ; (2) the 

 greater the difference in brightness between a standard colour and grey, 

 the more definite is the discrimination ; (3) the length of time required 

 to establish a successful discrimination habit between blue and red ; 



(4) the fact that whereas change of colour frequently appeared to pass 

 unnoticed, a marked change of brightness often caused confusion ; 



(5) the fact that green and yellow were not at all, or only with the 

 greatest difficulty, discriminated from white ; (6) the fact that during a 

 prolonged series increased confusion and uncertainty in discrimination 

 were shown from day to day. 



In favour of the view that dogs possess rudimentary colour dis- 

 crimination are (1) the decreased time for training required by each 

 new colour in the " approximate brightness value series " ; (2) the unique 



